on 2011 Oct 06 6:23 PM
Hi All,
When BWA is down for some reason, what is the course of action to
1) the ACR that needs to update the indexes?
2) the new requests that were not yet rolled up to BWA?
This question is similar to this thread
We know ACR will fail, But how do you handle the failed ACR and the rest of the data loads which might need the updated master data?
When the new requests do not roll up to BWA due to BWA downtime, these requests are not available for reporting. The users are basically reporting on the stale data until the requests are rolled-up. How do you handle this situation when it is critical for reporting?
And finally, how to handle the downtime through process chain? We planned to use the RSDDTREXHPAFAIL table entry to make a decision to continue with the BWA actions like roll-up etc. The other alternative was to use code similar to RSDDTREX_BIA_INITCHECK. I fail to see any information on handling connection failures in process chains.
During some situations, alert email does not come through as there is no entry made in the RSDDTREXHPAFAIL table, and there is no way we can pro-actively handle the situation.
Thanks,
Murali
Request clarification before answering.
Murali,
I have seen many people helped you in the past yet you haven't given any points to them - not that matters but just an fyi
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Tansu,
Thanks for pointing out that I was not courteous enough to thank the posters. My bad. I will award points and make it a practice.
Still I feel my question is unanswered. All the three posters were trying to say the same thing. To delete the cube and reindex and I understand that is the best option we have right now to ensure faster reporting of lastest information due to BWA downtime.
Regarding ACR, I was told that index update is the last step of ACR and if the connection is down, only the index will not be updated, but the BW SIDs are already updated. So I can ignore the ACR failure and continue with the BW loads as the MD is intact. This looks good. Only concern is will the later ACRs and roll-ups take care of the data that was missed in the earlier ACR. Still trying to find an answer.
From the reporting point, if reporting on the latest data is a absolute must, then the only option is to delete the index. But the issue is will very large cubes that will take a lot of effort to reindex.
I wished SAP provided an option to turn off the BWA index, both for reporting and loading purposes (we already have option for reporting, but not sure for loading), so that when BWA is down, I can turn off the index and the load process assumes that no index exist and no roll-up is needed and turns the request for reporting. Later when BWA is up, simply turning on the switch should restart the roll-up process for the new requests and make BWA in sync with BW. Can some one let me know if this option already exist (as I could not see one)? Other is SAP hearing us?
Awarding points is at the discretion of the poster. If he or she feels an answer was helpful, showed understanding of the question and gave some pointers to solution, then points should be awarded, as a matter of courtesy.
If the questions didn't really help, then giving no points seems appropriate.
I hope that clarifies things.
matt
Hi Murali,
There is no option as such for switching off the index for loading process. If you have BIA index created and due to some reason if it can not be rolled up then that request will not be available for reporting unless data roll up happens.
But if you delete the existing index and continue the loading then requests will be available for reporting and once BIA servers is up you can reindex the cube.
Regards,
Durgesh.
Murali,
If there is request in the cube and it's not available for reporting due to BWA downtime, I don't recommend any reporting available for the business. There is nothing more importing for the business/leadership to make accurate decisions based on trusted data. So when you say "steal data" it's not steal to me but it's incorrect data, so instead of providing a steal data i rather provide no data. Hope it makes sense. Having said that as many gurus said you need to delete the cubes from BWA.
I am curious why you have a long BWA outage though. I had 16x32GB blades, don't remember having 'planned' outage for months.
Cheers
Tansu
User | Count |
---|---|
53 | |
8 | |
6 | |
6 | |
5 | |
5 | |
5 | |
4 | |
4 | |
4 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.